摘要
针对燃煤电厂SCR脱硝系统入口NO_(x)浓度难以测量的问题,提出了基于改进鲸鱼算法(Improved Whale Optimization Algorithm,IWOA)优化双向长短时记忆神经网络(Bi-directional Long Short-Term Memory Neural Network,Bi-LSTM)的SCR入口NO_(x)浓度预测模型。利用LightGBM进行特征选择,运用最大时间周期的方法计算迟延时间;采用加入Relu层的Bi-LSTM神经网络提取时序特征,建立预测模型,并利用IWOA确定Bi-LSTM的最优超参数,最后与传统Bi-LSTM、LSTM、LightGBM预测模型进行对比验证。仿真结果表明,IWOA-Bi-LSTM模型的均方根误差、平均绝对百分比误差、平均绝对误差最小,能够实现对NO_(x)浓度的准确预测。
Aiming at the problem that it is difficult to measure the NO_(x)concentration at the inlet of the SCR denitrification system of coal-fired power plants,a prediction model of the NO_(x)concentration at the inlet of SCR based on the Bi-directional Long Short-Term Memory Neural Network(Bi-LSTM)optimized by the improved whale algorithm(IWOA)was proposed.We used LightGBM for feature selection and calculated the delay time with the maximum time period method.We used the Bi-LSTM neural network with the Relu layer to extract the timing features and established a prediction model,and used IWOA to determine the optimal hyperparameters of the Bi-LSTM.Finally,we compared the predication result with that of the traditional Bi-LSTM,LSTM,and LightGBM prediction models for validation.The simulation results show that the root mean square error,average absolute percentage error,and average absolute error of the IWOA-Bi-LSTM model are the smallest,which can achieve accurate prediction of NO_(x)concentration.
作者
姚宁
金秀章
李阳峰
YAO Ning;JIN Xiuzhang;LI Yangfeng(School of Control and Computer Engineering,North China Electric Power University,Baoding 071003,China)
出处
《华北电力大学学报（自然科学版）》
CAS
北大核心
2022年第6期76-83,共8页
Journal of North China Electric Power University：Natural Science Edition
